# News this Week

Science  15 Nov 2013:
Vol. 342, Issue 6160, pp. 782
1. # Around the World

1 - Tacloban, Philippines
Supertyphoon Poses Questions for Scientists
2 - Bucharest
Mining Project Plan Rejected
3 - Brussels
Europe Could Get New GM Maize
4 - Cold Spring Harbor, New York
New Biology Preprint Server
5 - New Delhi
India's Mars Mission Recovers
6 - Doha
Desert 'Greening' Experiment's First Results

## Tacloban, Philippines

### Supertyphoon Poses Questions for Scientists

Typhoon Haiyan, which raged across the Philippines last week packing wind speeds of up to 314 kph, is the most powerful tropical cyclone known to hit land. While Philippine authorities scramble to provide relief to survivors, scientists are pondering the storm's ferocity and seeking lessons for the future.

Geologist Jonathan Nott, of James Cook University in Cairns, Australia, notes that after starting out last week as a small storm, Haiyan underwent "very rapid intensification" on 6 and 7 November before making landfall just after 4 a.m. on 8 November.

Then it also moved "relatively quickly" across the islands. That sudden intensification left remote communities "precious little time to prepare for this event," says marine biologist Helen Yap of the University of the Philippines, Diliman. On the other hand, the storm's quick pace limited its impact. "The slower a storm system moves, the more wind damage it causes," Nott says. Hovering longer would also likely have soaked hillsides, raising the risk of landslides. Sorting out what this means for future mitigation efforts will take time, says Alfredo Mahar Lagmay, a geologist also at the University of the Philippines, Diliman. "Every disaster unfolds in a different way, and only after the disaster is over can you know its lessons," he says.

## Bucharest

### Mining Project Plan Rejected

Controversial plans to mine one of Europe's largest gold reserves are on hold. Romania's Rosia Montana region in Transylvania has been mined since pre-Roman times, and archaeologists say the site provides unique insights into millennia of mining technology (Science, 9 May 2003, p. 890).

On 11 November, a parliamentary commission unanimously rejected the draft law that would have allowed a Canadian company, Gabriel Resources, to begin mining. In recent months, thousands of protesters have marched in Romania to protest the bill, primarily because of plans to create a cyanide-laced tailings lake.

## Doha

### Desert 'Greening' Experiment's First Results

A project to "green" desert areas with a mix of technologies—producing food, biofuel, clean water, energy, and salt—reached a milestone last week. A pilot plant in Qatar, built by the Sahara Forest Project (SFP) and supported by Qatari fertilizer companies Yara International and Qafco, produced 75 kilograms of vegetables per square meter in three crops annually, comparable to commercial farms in Europe, while consuming only sunlight and seawater.

In SFP's greenhouse, fans blow hot desert air through a honeycombed curtain with salt water trickling down it to produce cool, moist conditions suitable for growing vegetables. Some of that moisture is recaptured by condensation to provide fresh water to irrigate the plants. Similar evaporative cooling is used to grow more crops outside, such as barley and arugula, and an onsite concentrated solar power plant provides electricity for the whole facility.

That a greenhouse with just 600 square meters of growing area produced such good yields suggests that a commercial plant could do even better, says SFP chief Joakim Hauge. SFP is now engaged in studies aimed at building a 20-hectare test facility near Aqaba in Jordan. http://scim.ag/greendesert

2. # Random Sample

## Oldest Slime Mats

While tromping across some of the world's oldest rocks, geobiologists led by Nora Noffke of Old Dominion University in Norfolk, Virginia, discovered what looked like rolled-up mats of sand several centimeters thick encased in rock. These turned out to be the oldest known "microbially induced sedimentary structures," distinctive shapes in sedimentary rock formed by layers of now-gone slime.

All manner of microbes can live layered on the bottom of shallow waters. Such microbial layers bind sand grains to form sand-laden mats, so that when, say, a strong current rips off a bit of mat and rolls it up, the sand form can be found embedded in rock eons later. In the December issue of Astrobiology, the scientists report finding a fresh mat fragment on a North Carolina barrier island (right) and examples lingering in 2.9-billion-year-old rock in southern Africa (middle) and in 3.5-billion-year-old rock in Western Australia (left). No vestige of life, micro- or megascopic, older than the Australian example has been found.

## Sex and the Single Sand Flea

The sand flea Tunga penetrans spends half of its life in the ground—and the other half in a living host, such as a human foot. Infestations—widespread in South America and sub-Saharan Africa—can produce a painful condition called tungiasis that makes it difficult to walk and can lead to other infections. But the disease gets scant attention from health workers or researchers, who know relatively little about the creature's life cycle—not even where (in the ground or in a host) it has sex and fertilizes its eggs.

So when Marlene Thielecke, a Ph.D. student at Charité University Medicine in Berlin who was studying tungiasis in Madagascar, discovered that she was hosting the parasite, she let her own body become a laboratory to study the creature.

The immature female burrows into a host's skin where, over 2 weeks, it swells to up to 10 mm across. She soon begins expelling eggs, then dies after 4 to 6 weeks. The only way to get rid of the parasites is to dig them out.

Thielecke took regular photographs of her infestation. The flea grew normally, but it wasn't laying eggs. It was also oddly long-lived—it was still alive after 2 months (at which point she extracted it).

The likely explanation? The flea was never fertilized, Thielecke and her supervisor, tungiasis expert Hermann Feldmeier, concluded in a paper published this month in Travel Medicine and Infectious Disease. On finding the flea, Thielecke had taken precautions, wearing socks and closed shoes to ward off other fleas—including potentially fertilizing males. This suggests an answer to the longstanding puzzle of whether fleas are fertilized before or after they embed. http://scim.ag/fleafert

## By the Numbers

70%—Percent of U.S. research universities that say the sequester reduced research money for faculty and slowed ongoing research projects, according to a poll released this week.

600,000—Bat mortality from wind turbines in the contiguous United States in 2012, based on an estimate in BioScience.

4. # The Man Who Bottled Evolution

1. Elizabeth Pennisi

Richard Lenski's 25-year experiment in bacterial evolution shows no signs of running out of surprises about how mutation and selection shape living things.

EAST LANSING, MICHIGAN—When most biologists want to understand how evolution unfolds, they look for clues in the fossil record or the natural world. Richard Lenski simply walks across his Michigan State University lab to his freezers. There, stored in 4000 vials, are bacteria dating back to 1988. That was the year Lenski started a simple but radical experiment. He put samples of Escherichia coli into a sugar solution, stoppered the flasks, and waited to see what would happen. It was a study with no defined endpoint, so risky that he didn't try very hard to get outside funding for it.

After 25 years and 58,000 bacterial generations, Lenski's bacteria are still growing, mutating, and evolving. They are proving as critical to understanding the workings of evolution as classic paleontology studies such as Stephen Jay Gould's research on the pace of change in mollusks. Lenski's humble E. coli have shown, among other things, how multiple small mutations can prepare the ground for a major change; how new species can arise and diverge; and that Gould was mistaken when he claimed that, given a second chance, evolution would likely take a completely different course. Most recently, the colonies have demonstrated that, contrary to what many biologists thought, evolution never comes to a stop, even in an unchanging environment. The work is "an absolutely magnificent achievement," says Douglas Futuyma, an evolutionary biologist at Stony Brook University in New York.

Other researchers have done experimental evolution, setting up populations of insects, yeast, and even fish in the lab and in controlled field conditions, and subjecting the organisms to a particular environmental stress for relatively short periods. But Lenski's long-term experiment "is just orders of magnitude beyond what anyone else has done," Futuyma says.

The project's quarter-century has witnessed the rise of bioinformatics and the birth of whole-genome sequencing, and Lenski has taken advantage of both technologies to glean new insights. Generations of students have tended and analyzed the microbes, and the project sparked a memorable conflict between Lenski and creationists. Fifteen years ago, he almost abandoned it for digital models of evolution, then reconsidered—and was vindicated when his bacteria took one of their most dramatic evolutionary leaps. As Scott Edwards, an evolutionary biologist at Harvard University, said at a June evolution meeting,* "the principles of evolution that Rich has uncovered have touched all of us and caused us to look at evolution in a new way."

## Hard data

Lenski, 57, has been a MacArthur Fellow, a member of several editorial boards, and a society president; he belongs to the National Academy of Sciences and served on committees that evaluated genetically modified organisms and the investigation of the 2001 anthrax attacks. In August, he became an avid tweeter and started his own blog, Telliamed Revisited, named after an 18th century book that stressed the importance of understanding the world through observations and not religious dogma. He has collected old scientific texts and baseball statistics, although these days, he says, "my granddaughter is my current hobby."

But one thing will always capture his attention. Show up at his door with new data, and he's "like a 5-year-old at Christmas," says graduate student Michael Wiser. That hunger for hard data about evolution was what drew him to bacteria in the first place. Lenski never had a course in microbiology. As a graduate student at the University of North Carolina, Chapel Hill, he studied the ecology of ground beetles—work he found interesting but limited. "[I]t was difficult to imagine feasible experiments that would really test the scientific ideas that most excited me," he wrote in an August blog. For his postdoctoral training in 1982, he switched gears and joined the lab of Bruce Levin, one of the few researchers doing experimental evolution in microbes. They studied the interactions between bacteria and viruses called bacteriophages. But their systems were still too complex to get at the question Lenski most wanted to answer: Was Gould right about evolution's irreproducible nature, or would evolution often repeat itself if given a second chance?

So after starting his own lab at the University of California, Irvine, Lenski used some of his own money, from a National Science Foundation (NSF) Presidential Young Investigator Award, to set up a much simpler system using just bacteria. It was "like a physics experiment [where] you try to strip [out] all the complications so you can isolate the phenomenon that you want to describe," explains Lenski postdoc Noah Ribeck, a physicist.

In 1988, Lenski placed identical E. coli populations into 12 flasks filled with a liquid containing nutrients and 25 milligrams of glucose per liter. Every 24 hours, with constant stirring and a comfortable 37°C environment, the bacteria multiplied explosively, depleting the sugar. Each day, 1% of the bacteria were transferred to a new glucose-laden flask. Every 75 days—about 500 generations for E. coli—Lenski's team froze some of the bacteria for future studies and as backups should the experiment become contaminated. This protocol has continued day in and day out, weekends and holidays, virtually uninterrupted for 25 years. (The cultures were temporarily frozen during one winter break when the campus was deserted and when Lenski moved his lab from California to Michigan.)

Setting up the protocol and sticking with it "took a certain spirit and vision," Levin says. Early on, skeptics argued that evolution in bacteria, particularly bacteria growing in a stable, comfortable environment, wouldn't reveal much about how the process unfolds in nature. Yet, as they monitored their cultures, Lenski and his colleagues saw what looked very much like an accelerated process of evolutionary change. By using bacteria instead of a slower-growing organism, Lenski compressed time: The E. coli go through 6.6 generations in a day, compared with more than a year for the same number of generations in mice. In 25 years, he's seen the equivalent of a million years of evolution in humans. With that span, "you can examine a whole range of ideas that there hasn't been an opportunity to look at before," says Graham Bell, an evolutionary biologist at McGill University in Montreal, Canada.

By thawing out frozen vials from generations past, Lenski can go back in time to look at intermediate stages of evolutionary change. "It's like the perfect fossil record," says Douglas Emlen, an evolutionary biologist at the University of Montana in Missoula. "They can pinpoint exactly when [a trait] arose." And Lenski can restart the experiment with those ancestral bacteria to see if history repeats itself. "A graduate student who is 22 today can study populations that come from before they were born and use techniques that haven't existed before," says Christopher Marx, a microbiologist at Harvard University and a former Lenski postdoc.

For the first decade, things hummed along. The investigators monitored the bacteria's fitness by measuring how fast they could multiply relative to their ancestors. At first, their fitness rapidly increased, but then the improvement slowed. Over the whole experiment to date, fitness has improved by an average of 70%—meaning the most recent descendants undergo 1.7 doublings in the time it took the original microbes to double once.

All 12 lines improved by about the same amount, showing that, broadly speaking, evolution is reproducible (Science, 25 June 1999, p. 2108). But some lines improved faster than others, and genetic analyses of the strains showed that they had taken different evolutionary paths. For example, six lines developed defects in DNA repair, and instead of dying out, began sustaining higher mutation rates than their counterparts. "Historically, which mutations arise early changes the subsequent evolution," explains Janette Boughman, an evolutionary biologist at Michigan State University. "But you really get to the same endpoint": better fitness in sugar solutions.

Other striking differences emerged among the flasks. At generation 6500, about 3 years into the experiment, two types of E. coli evolved in one of the flasks: one that made small colonies consisting of relatively small cells and one that made large colonies, with large cells. Lenski expected that eventually one type would take over and the other would disappear, or both would be ousted by a bacterium with an even more beneficial mutation. But to his surprise, both types have persisted, creating an ecosystem in which competition and other interactions between the colony types allowed both to be viable. "He created his own Galápagos Islands," Marx says.

## Going digital

Having done so, Lenski almost gave up on the flask experiment. He had caught a glimpse of an experimental evolution system that promised to be even simpler and faster. "He's a shiny object researcher," says former student Paul Turner, now an evolutionary biologist at Yale University. "If there's something interesting to pursue, he will try to pursue it." During a squash game, a physics colleague invited Lenski to a seminar by California Institute of Technology physicist Christoph Adami. With his graduate student Charles Ofria, Adami had developed software that allowed self-replicating computer programs, aka digital organisms, to compete with one another for processing power and evolve new functions. Lenski was entranced. "One of the first data slides [Adami] put up looked like data from my long-term experiment," he recalls.

In 1998, Adami sent Lenski an advance copy of his book Introduction to Artificial Life (Science, 8 May 1998, p. 849). Lenski stayed up for 2 nights straight reading about the program and trying out simulations he had designed himself. The digital organisms replicated thousands of times faster than his microbes, and their "mutations" could be tracked in more detail. For the next 6 years, he focused as much on digital evolution as on bacteria.

In 2003, he teamed up with Ofria and Adami, both of whom are now at Michigan State, and with Robert T. Pennock, a philosopher there, to follow mutation by mutation how computer programs that initially could do no more than replicate evolved the ability to perform complex operations, such as checking whether one numerical string equals another. They found that many earlier mutations—some of them deleterious in the short term—had to accumulate before a final "enabling" mutation conferred the new trait. The work demonstrated that complex traits, such as the vertebrate eye, likely come about through a series of intermediate steps that open the way for future adaptation.

Lenski's fascination with digital life hasn't faded. In 2010, he, Ofria, and others got funding for BEACON, an NSF Center for the Study of Evolution in Action, which includes 400 investigators and students from five U.S. universities. They collaborate on studies of evolution as it occurs in both real and digital organisms.

Fortunately, he did not abandon his first love. The weekend that Lenski discovered digital organisms, he had told his wife he might shut down the long-term experiment. Only her nudging convinced him to keep it going—and in January 2003, the value of doing so became clear, when the bacteria served up yet another surprise. "Digital organisms have the advantage that you can get complete information about pretty much all aspects of an experiment," Lenski explains. "The bacteria, though, are a lot more complex and so they have a lot more tricks up their sleeves, a lot more potential to evolve in ways that one cannot anticipate."

One morning, Lenski and his colleagues noticed that the medium in one flask had grown turbid, a sign that it was unusually thick with bacteria. They suspected contamination but could not confirm it, so they dug out the most recent frozen sample of Ara-3, as that population was called, and restarted it. Three weeks later, the turbidity reappeared. This time, they thoroughly tested the culture and ruled out contamination. By growing the Ara-3 bacteria on different types of media, they discovered that the bacteria in that flask had evolved a new way to nourish themselves. Instead of relying on scarce glucose, they drew on a different energy source in their medium, citrate, which enabled them to reach much higher densities than in other flasks. "This was the biggest event in the entire E. coli experiment," Adami says. "To have a complex new function develop seemingly from scratch is a big deal and quite remarkable."

It fell to graduate student Zachary Blount to figure out what happened. Was a single mutation responsible? And if so, why hadn't this ability appeared earlier, since the bacteria had already had plenty of time to experience mutations in every possible gene? Or were multiple mutations required, and if so, when and in what order did they arise? To find answers, Blount rewound the tape of life, unfreezing earlier generations and letting them evolve again in the same medium. Citrate use emerged in four of 72 cultures, always in ones derived from more recently frozen samples. That suggested that multiple mutations underlie the trait. Populations from those later generations had already acquired the predisposing mutations and so could more readily take the next step.

Blount and Lenski's results, published in 2008, drew lots of attention, and not just from the media and scientific community. On his website, Conservapedia, conservative Christian Andrew Schlafly questioned the conclusion that the bacteria had evolved this striking new ability and requested Lenski's original data. Lenski referred him to the original paper, but Schlafly persisted. Lenski responded with a sharply worded response that went viral. "He immediately became a rock star of science," Adami says.

Meanwhile, the citrate work continued. The cost of sequencing microbial genomes dropped dramatically, enabling Blount to look at the specific genetic changes involved. He found that citrate users had undergone a duplication of a 2933-base piece of DNA that activated an otherwise silent gene coding for a citrate transporter. Subsequent genetic changes then tuned up the efficiency of that transporter, he and Lenski reported in 2012.

These citrate users are enabling Blount, now a postdoc in the lab, and Lenski to look at another aspect of evolution: the formation of new species. The usual test of separate species is that they are unable to interbreed successfully, a criterion that can't be applied to bacteria because they don't mate. But because one of E. coli's defining characteristics is the inability to use citrate for energy in the presence of oxygen, the citrate-consuming bacteria could be seen as a new species. And they may even meet the traditional definition. Researchers can't interbreed bacteria, but they can mix the genomes of separate strains. Bacteria that thrive on citrate do poorly on glucose, and melding the citrate users with the parent strain produces a less fit hybrid, Lenski reported at the June evolution meeting.

## Evolving endlessly

Lenski no longer thinks about ending his experiment. "It's become more and more apparent that it's not good to just do short-term experiments," he says. "Any microbiology lab with enough people should be thinking about doing a 20-year experiment." Funders agree. The project, once rejected by the National Institutes of Health, now has ongoing NSF support: a 10-year Long Term Research in Environmental Biology grant.

And it is continuing to pay off, Lenski, Wiser, Ribeck, and their colleagues report online this week in Science (http://scim.ag/MWiser). Researchers have long assumed that when organisms encounter a new environment, they will adapt very quickly at first, then, as long as conditions are stable, ultimately reach an adapted state. At that point—a "fitness peak"—adaptive evolution should virtually stop. Wiser tested the fitness of all 12 populations at 41 time points during their evolution, traveling back in time with the frozen samples. At 10,000 generations—about 5 years—it seemed the bacteria were reaching that fitness peak. But now, after 50,000 generations, the improvement has slowed down but not leveled off as expected, Lenski's team reports. "The notion of a fitness peak is more elusive than I anticipated," Lenski says. "I think fitness may well continue to increase for a million years." Evolution is endless, it seems, even in a stable environment. "That's a profound insight," Boughman says.

It's also an encouraging portent for the future of Lenski's cultures, which are now passing 58,500 generations. "If you asked me 20 years ago, I thought [the researchers] were running out of new things to learn," Levin says. Now, "I think they should go on indefinitely." Lenski agrees. During his presidential address at the recent evolution meeting, he made a plea: "If you know anyone who would like to endow a million-year experiment, have them get in touch with me."

• * Evolution 2013, Snowbird, Utah, 21 to 25 June.

5. # Turning Up the Light

1. Robert F. Service

Photovoltaic materials called perovskites work wonders in the lab, but will they shine as commercial technology?

Plot the progress of different types of solar cells, and one line stands out. For decades, almost all solar technologies—such as panels made with wafers of crystalline silicon or thin films of cadmium telluride—have made slow, steady progress. A technology that converts 14% of the energy in sunlight to electricity nudges up to 14.1% 2 years later, and so on.

But a new contender, solar cells built with complex crystalline materials called perovskites, is leaping ahead. It tiptoed onto the scene in 2009 with cells that were 3.8% efficient—a ho-hum result when top silicon cells in labs were notching 25%. By the end of 2011, however, that efficiency had nearly doubled to 6.5%. Last year, it climbed to 10%. In 2013, the new cells have hit 15%, surpassing some alternatives that have had decades to mature. "It's amazing," says David Cahen, a materials scientist at the Weizmann Institute of Science in Rehovot, Israel. "We have never seen anything like this in the solar cell community."

The news gets better. Perovskites are made from readily available materials. Unlike some kinds of solar cells, they're cheap and easy to produce. Experts think they have plenty of room for improvement and that their efficiencies could top 20% within the next year. And perhaps best of all, perovskite solar cells have the potential to be integrated with silicon panels, creating tandem cells with efficiencies of 30% or more. "There is a wave of excitement, and it's spreading," says Michael McGehee, a materials scientist at Stanford University in California.

Perovskite cells are a long way from a commercial debut: So far, most are no bigger than a postage stamp, produce just milliamps of electricity, and dissolve on exposure to air. The best of them also contain lead, an environmental toxin. But researchers are scrambling to surmount these challenges. To date, only a couple of dozen papers on the new cells have been published, most by a handful of groups. But that's likely to change quickly. "People are pouring into [the field]," McGehee says. From his extensive contacts in the area, McGehee estimates that more than 100 groups around the world are now working on perovskites.

"It's very competitive," says Michael Grätzel, a chemist at the Swiss Federal Institute of Technology in Lausanne, Switzerland. "The battle is on. One has to move very quickly." Yang Yang, a solar cell expert at the University of California, Los Angeles, agrees. "I don't have enough time to sleep," Yang says.

## The right mix

Perovskites have been under the noses of solar cell makers for more than a century. A Russian mineralogist discovered the first, natural version in 1839. Today, hundreds are known. The category simply refers to materials that adopt the same type of crystal structure with atoms arranged in octahedra that are connected at the corners in all three spatial directions. The solar cell perovskites are semiconductors. But other family members run the gamut from conductors to insulators, the most famous being the high-temperature copper oxide superconductors that burst onto the scene in 1986.

In the 1990s, David Mitzi, a physicist at IBM's Thomas J. Watson Research Center in Yorktown Heights, New York, made thin-film transistors and light-emitting diodes from a semiconducting perovskite. The devices worked. But even though many light-emitting materials also make good light absorbers—and thus potentially good photovoltaics—Mitzi decided his perovskites were too unstable for use in solar cells, which must survive decades to be commercially viable.

Nearly a decade later, Tsutomu Miyasaka took a first step toward solving that problem. Miyasaka, a chemist at Toin University of Yokohama in Japan, and colleagues were working on photovoltaics called dye-sensitized solar cells (DSSCs). Unlike conventional silicon solar cells, DSSCs consist of a blend of organic light-absorbing dyes coating tiny inorganic particles such as titanium dioxide (TiO2), which in turn are surrounded by a charge-conducting electrolyte. In standard DSSCs, when a dye molecule absorbs a photon, the light boosts the energy of one of the electrons in the dye, enabling it to jump onto a TiO2 particle. From there, it skips from particle to particle until it reaches an electrode, where it's collected and sent through a circuit to do work. Meanwhile, another electron jumps from the electrolyte to the dye to restore it to its original state.

The trouble, says Grätzel, whose team invented DSSCs in 1991, is that the dyes don't absorb all the light that hits them, reducing the cells' efficiency. Hoping to do better, Miyasaka turned to a perovskite. He says that it took one of his students 2 years to find a recipe to make the material sturdy enough for a brief demonstration. In the 6 May 2009 issue of the Journal of the American Chemical Society, they reported building cells that, in place of the dye, contained a thin layer of a light-absorbing perovskite and were 3.8% efficient. Unfortunately, the cells also contained a liquid electrolyte that dissolved the perovskite within minutes, causing them to fail.

Grätzel took the next step, with Nam-Gyu Park and colleagues at Sungkyunkwan University in Suwon, Korea. On 21 August 2012 in Scientific Reports, they reported that they had replaced the liquid electrolyte in a perovskite-containing cell with a solid version and achieved efficiency close to 10%. Now things were starting to get interesting.

Grätzel then looked for better ways to grow the perovskites. Previously, his group and others had mixed precursor compounds in a solvent and heated it to drive off the solvent and crystallize the perovskite. But the approach produced perovskite layers that widely varied in thickness. Hoping to control the process better, they came up with a two-step recipe using two different solvents. In the 18 July issue of Nature, they reported that the process yielded more even coatings for their TiO2 and an efficiency of 15%.

Meanwhile, at the University of Oxford in the United Kingdom, Henry Snaith, one of Grätzel's former postdoctoral assistants, took the next step toward all-perovskite cells. Snaith's team wondered whether the TiO2 played a crucial role in the solid-electrolyte cells. They tried replacing the semiconducting TiO2 with a porous insulating material made from aluminum oxide (Al2O3), thinking that removing the semiconductor would kill the performance of their cells. Instead, the cells worked better than before.

The researchers realized that the perovskite was acting not only as a light absorber, but as a charge-carrying semiconductor as well. In that case, they reasoned, why bother with TiO2 or Al2O3? In the 19 September issue of Nature, Snaith and his colleagues reported that they simply grew a thin film of their perovskite, with a couple of additional layers sandwiched around it to help ferry charges to the electrodes, and matched the Gratzel team's efficiency of 15%.

The progress has been so rapid that Snaith says, "I think it could be feasible to get physically addicted to breakthrough results."

## Better and better

So why are perovskite solar cells surging ahead when other technologies have been struggling to break 12%? A good part of the answer, Cahen says, is that the perovskites have near-perfect crystallinity. That's a trait shared by today's top solar cell materials, including gallium arsenide and crystalline silicon.

In second-tier cell materials, this crystalline arrangement is typically riddled with defects. When electric charges whizzing through the crystal run into those defects, they stop cold and often give up their excess energy. Growing defect-free crystals normally requires ultrahigh temperatures or multimillion-dollar machines, such as molecular beam epitaxy. But perovskites can be grown at 80°C and simply precipitate out of solution in near-perfect form. "It's a little bit of a dream come true," Cahen says.

In the 18 October issue of Science, separate teams that included Snaith and Grätzel reported a consequence of that perfection: Perovskites are excellent at allowing excited electric charges to travel long distances through the material. That property, known as the carrier diffusion length, is critical for all solar cells. It measures how far an electron might travel before it meets up with a positively charged electron vacancy, or hole, and drops into it. In the process, the electron gives up the excess energy that it got from the photon of sunlight, generating heat instead of electricity.

Organic solar cells typically have a diffusion length of about 10 nanometers. In perovskites, by contrast, carriers can travel as much as a micrometer, 100 times as far. "The upshot is, you can collect charges over longer distances," Grätzel says—thus making it more likely that the charges will wind up as useful current.

Perovskites have another highly valued property that boils down to how effective they are at generating an electric voltage. In crystalline silicon solar cells, for example, it takes photons with energies of at least 1.1 electron volts (eV) to kick an electron out of the tight grip of a silicon atom and become freely conducting. By the time those electrons reach an electrode and are dumped into a circuit, their voltage drops to 0.7 (eV), a loss of just 0.4 eV—part of the reason silicon is so commercially successful. For traditional DSSCs and organic solar cells, these losses usually amount to 0.7 to 0.8 eV. Perovskites, however, match the commercial champs with losses of only 0.4 eV. The new solar cells also do a decent job on a third solar cell metric, known as the fill factor, which measures how much power a solar cell produces relative to its theoretical maximum.

"For perovskites, these three parameters are all very good," Yang says. "It's just what we want." Yang's group has spent much of the past 10 years pioneering work on organic solar cells to raise their performance close to 11%. But since they've started tinkering with perovskites, "we've achieved about 13% in 5 months," Yang says.

Such overnight successes have made researchers confident that they can do even better. Grätzel notes that top-flight solar cells have a fill factor of about 80%, out of a theoretical maximum of 90%. But most perovskite cells today are between 60% and 70%. Likewise, early results suggest there is room for improving the amount of current the cells generate as well. As a result, McGehee says, he expects the overall efficiency of perovskite solar cells to keep rising. "We may see 20% before the end of the year," he says.

They may go even further with help from today's champion: silicon. Commercial silicon cells already deliver 17% to 23% efficiency. "Making the silicon industry obsolete is very tough," McGehee says. Alternative technologies, such as thin films of copper indium gallium selenide, use far less material and thus are cheaper to make. But they haven't been able to knock silicon from its perch because they are less efficient. Consumers need more panels to generate the same amount of electricity and thus must pay higher installation costs.

Perovskites, by contrast, might literally piggyback off silicon's success. Perovskites are better than silicon at absorbing higher energy blue and green photons, while silicon excels at snagging lower energy red and infrared photons. And because perovskites can be grown at temperatures below the melting point of glass, engineers may be able to layer them directly onto the top glass coating of a silicon cell. That strategy, McGehee suspects, could produce low-cost tandem cells with about 30% efficiency. No one has done it yet, but McGehee says he expects many researchers are trying.

All of this comes as a much-needed boost for the solar cell industry, which has struggled through a brutal commercial retrenchment in recent years as falling prices for solar cells have caused numerous companies to go bankrupt. Venture capital firms and even scientific funding agencies were getting cold feet about supporting research in slow-moving areas such as organic photovoltaics and DSSCs, Grätzel says. "The mood was extremely grim in the PV community." So perovskites couldn't have come at a better time, he adds. "It was the shot in the arm that we needed."

Even so, perovskite solar cells have a long way to go before they will be ready for the marketplace. For starters, Cahen says, most cells being produced in labs today are tiny, just centimeters on a side. By contrast, silicon panels are meters across. "It's very difficult to grow large continuous films" of perovskites, Cahen says. Nor have researchers solved the durability problem. Perovskite cells are extremely sensitive to oxygen, which reacts to break down the crystal structure, and to water vapor, which dissolves the saltlike perovskites. Even worse, the lead in today's best perovskites could leach out of the solar panel onto rooftops or the soil below.

"There are problems here," Cahen says. "At the moment, I'm an optimist, and a believer in materials research." Between the promise of perovskites and the multiple challenges that still need to be solved, "this field is going to be extremely active for the next few years," Grätzel says. And with the global solar cell market still worth nearly \$50 billion a year, researchers have every incentive to keep the progress curve rising.